“The structure of the Type III secretion system export gate with CdsO, an ATPase lever arm”

Type III protein secretion systems (T3SS) deliver effector proteins from the Gram-negative bacterial cytoplasm into a eukaryotic host cell through a syringe-like, multi-protein nanomachine. Cytosolic components of T3SS include a portion of the export apparatus, which traverses the inner membrane and features the opening of the secretion channel, and the sorting complex for substrate recognition and for providing the energetics required for protein secretion. Two components critical for efficient effector export are the export gate protein and the ATPase, which are proposed to be linked by the central stalk protein of the ATPase. We present the structure of the soluble export gate homo-nonamer, CdsV, in complex with the central stalk protein, CdsO, of its cognate ATPase, both derived from Chlamydia pneumoniae. This structure defines the interface between these essential T3S proteins and reveals that CdsO engages the periphery of the export gate that may allow the ATPase to catalyze an opening between export gate subunits to allow cargo to enter the export apparatus. We also demonstrate through structure-based mutagenesis of the homologous export gate in Pseudomonas aeruginosa that mutation of this interface disrupts effector secretion. These results provide novel insights into the molecular mechanisms governing active substrate recognition and translocation through a T3SS.


Introduction
Bacterial pathogens secrete toxins and other effectors to promote virulence by subverting host processes and defenses through the evolution of specialized secretion systems (type I to type the export apparatus and links the inner membrane protein complex to the ATPase (SctN) [4,15,16]. The export gate is known to undergo an opening and closing of the cleft between subdomains 2 and 4 (the SD2-4 cleft) and closing of this cleft promotes substrate release [17][18][19]. The export gate does not directly engage the sorting complex, but is linked to the ATPase by SctO [8], an~140 Å coiled-coil that is structurally similar to the central stalk proteins of the rotary ATPases-the F 1 -ATPase γ-subunit and the V 1 -ATPase D subunit [7,20,21]. SctO is essential for substrate secretion [22][23][24]. The T3S ATPase itself is structurally related to the F-and V-type ATPases and has been proposed to function with a similar rotary catalytic mechanism wherein a coiled-coil subunit (SctO) engages the asymmetric pore of the homo-hexameric ATPase and SctO rotates during ATP hydrolysis cycles, shifting interactions to neighboring ATPase subunits coincident with ATP hydrolysis [1,25]. In T3SS, SctO is the key link between the export gate and the ATPase and is poised to transmit mechanical force between the ATPase and the export gate. A recent cryo-EM structure of the T3SS ATPase:central stalk complex from E. coli (EscN:EscO) revealed a single EscO extending away from EscN at an~70˚angle, and comparison of the EscN homohexamer and the EscO-bound structures suggests a rotary catalytic mechanism similar to that observed for the F-and V-type ATPases, in which EscO rotates during ATPase catalysis [25].
Presented in this manuscript are structures of the C-terminal region of the export gate from Chlamydia pneumoniae (CdsV), both in an unliganded form and when bound to residues 25-110 of the Chlamydial SctO (CdsO). These structures show that CdsO engages CdsV in a cleft between adjacent subunits and influences the configuration of the SD2-4 cleft, thus revealing how the ATPase may control substrate release by rotating CdsO.

Structure of CdsVc
We determined the crystal structure of the CdsVc (CdsV C-terminal region) homo-nonameric ring assembly and refined the structure to 2.8 Å (Table 1; Fig 1B and 1C; PDB 6WA6). The crystallized protein contains residues 345-710 of CdsV from Chlamydia pneumoniae; several N-and C-terminal residues from most monomers could not be resolved from electron density difference maps. S1 Table and Table 1 describe the statistics of the refined models and the contents, or completeness, of those models. The amino terminal region, approximately residues 1-345 of CdsV and other export gate homologs, are predicted to contain 6 transmembrane helices, which anchor CdsV to the inner membrane. CdsV C displays the same fold as homologs MxiA from Shigella flexneri, InvA from Salmonella typhimurium, and FlhA monomers from S. typhimurium, Bacillus subtilis, and Helicobacter pylori, with four distinct subdomains (subdomains 1-4) ( Fig 1C) [17,[26][27][28][29][30]. Monomers of CdsV C align with RMS deviations of 0.26-1.68 Å; the primary differences across the nine subunits exist in subdomains 2 and 4, and in particular, the cleft formed between subdomain 4 of neighboring CdsV protomers, as evidenced by the high B-factors observed in the structure (Fig 1B, S6A Fig). The closed, planar ring is stabilized by the buried surface area between subdomains 1 and 3, with an average total interaction area of 1127 Å 2 , as well as several salt bridges and hydrogen bonds between conserved residues (S2 and S3 Figs), as noted for MxiA [26]. The CdsV C nonamer has an inner pore diameter of~60 Å, with the total diameter of the ring~170 Å.
Residues lining the inner surface of the ring, which correspond to subdomain 3, are highly evolutionarily conserved across prokaryotes with T3S injectisome machinery (S2 and S4 Figs) and flagellar T3S, while residues along the outside surface (subdomain 2) are highly variable. This suggests a conserved functional role, such as substrate secretion, for residues lining the pore, while divergence of the outer surface may allow the export gate platform to form multiple

PLOS PATHOGENS
Structure of the Type III secretion system export gate with CdsO species-specific interactions. Indeed, deletion of subdomain 2 of MxiA did not abolish effector secretion, but did impact secretion of translocon components [26]. As was shown for MxiA, conserved residues lining the CdsV C pore also include several lysines and arginines (S4 and S5 Figs), which are critical for secretion [26].

Structure of the CdsV C :CdsO complex and oligomer assembly
We also determined the crystal structure of CdsV C in complex with a portion of CdsO and refined the structure to 4.6 Å (Table 1; Fig 2; PDB 6WA9). The CdsO protein was truncated to residues 25-110 to facilitate crystallization of the complex. Although determined at a relatively low resolution, high-resolution structures of CdsV C and the 84% identical CdsO from Chlamydia trachomatis, for which a high resolution structure exists [31], greatly simplified

PLOS PATHOGENS
Structure of the Type III secretion system export gate with CdsO structure determination and interpretation. This structure defines the structural organization of the export gate bound to SctO. Several residues on both termini of CdsO  could not be modeled, due to the limited resolution and poorly resolved electron density in the area (S1 Table). Most notably, we observed CdsO 25-110 positioned within a large cleft at the interface between two CdsV C protomers, specifically, between subdomain 4 of adjacent subunits (the SD4-4 cleft, see Fig 2A). In our crystals, a 1:1 stoichiometry between CdsV C and CdsO 25-110 is observed, and saturation of CdsV C in this way likely aided crystallization. CdsO  binding is mediated largely through electrostatic interactions and stabilized by the buried surface area of each face of the CdsO 25-110 coiled-coil with one side of the CdsV C monomer (Fig 2B and  2C). The interacting residues on CdsV are fairly well conserved (S2 and S4 Figs and Fig 2C), suggesting that a similar interaction may occur in other homologs. The average total interaction area of one CdsO 25-110 with a CdsV C dimer is~900 Å 2 .
Large-scale architectural rearrangements were not observed in CdsV C upon binding CdsO   (Fig 2B, S6A Fig); instead, small conformational differences were identified primarily within subdomain 4. Helices 12 and 13 of CdsV C , connected by an extended loop, were displaced by an average of 3.9 Å and 9.5˚when bound to CdsO  , which draws subdomain 4 further into the SD4-4 cleft to stabilize CdsO  . Comparison of CdsV C with CdsV C : CdsO    [32]. These binding clefts thus appear functionally linked, wherein binding at one site may promote binding at the other site. Helices 12 and 13 of subdomain 4 shift considerably into the region that forms the SD2-4 cleft when in the "closed" form (S7E and S7F Fig), which would occlude or restrict binding of CdsO. These regions are somewhat flexible in the absence of ligands, are the sites of the greatest structural differences between protomers, and the locations of missing electron density or high B-factors in both CdsV and MxiA [26]. Density for the loop connecting helices 12 and 13 is observed in the CdsV C :CdsO 25-110 complex but is unresolved or only partially resolved in the structure of CdsV C alone, despite this structure being at higher resolution than the CdsV C :CdsO 25-110 complex, further supporting the idea that these clefts are flexible and become more stable when ligands are bound.
In FlhA, residues 621-641 form a helix (part of subdomain 4) that shifts position to allow for substrate binding ( Fig 3F) [18,28]. In Chlamydia, this helix is shorter, as it is interrupted by Pro 656 and Pro 658, although these helix-breaking residues are not conserved in T3SS from other organisms (Fig 3F and S4 Fig). These structural changes would significantly alter the

PLOS PATHOGENS
Structure of the Type III secretion system export gate with CdsO

PLOS PATHOGENS
Structure of the Type III secretion system export gate with CdsO chaperone-binding site identified in FlhA such that Chlamydia may use a somewhat different substrate recognition strategy, likely still involving the large pocket that remains accessible in CdsV.
CdsO shares the coiled-coil motif with homologs from other injectisome and flagellar systems, and an existing structure for CdsO from Chlamydia trachomatis (PDB 3K29; [31]) was used for molecular replacement. The complex revealed the binding site of CdsO 25-110 , although, due to the limited resolution of the structure, only residues~39-85 could be modeled per coiled-coil (S1 Table). The CdsO 25-110 structure exhibits a key structural constraint. The loop connecting the two helices of the coiled-coil has a small, hydrophobic or uncharged residue midway between the helices, with its sidechain pointed parallel with the long axis of the coiled-coil (Thr 67 in C. pneumoniae CdsO and C. trachomatis CdsO; Val 64 in YscO from Vibrio parahaemolyticus; Gly 58 in FliJ). The presence of a small residue with its sidechain pointed along the coiled-coil results in a backbone-mediated interaction between CdsO  and CdsV (Fig 2C). This interaction is formed between the last beta strand of CdsVc and the loop of CdsO  . Similar to the other YscO-like proteins, the two helices exhibit amphipathic packing of the sidechains central to the monomeric coiled-coil. Despite the common helixloop-helix motif, YscO-like proteins display significant divergence in primary sequence (S8A and S8B Fig) and in protein size, as YscO-like proteins vary in length by as many as 40 residues. However, a commonality of the injectisome T3S SctOs is the conserved structure that is able to dock within the appropriate SctV and interact with the conserved sites at the base and sides of the SD4-4 cleft (Fig 2C). This interaction immobilizes both helices and the short loop between them.
To further evaluate the CdsV C : CdsO 25-110 complex in the context of the full injectisome, the CdsO 25-110 structure was manually extended to contain residues 1-162 of the 168 residues of full-length CdsO from C. pneumoniae, using the C. trachomatis CdsO as a template (S9A

Mutations that disrupt the CdsV:CdsO interaction decrease secretion when introduced into PcrD in Pseudomonas
To functionally assess both the importance of the CdsV:CdsO interaction and its conservation in other T3SS, two structure-guided CdsV mutations were designed to disrupt the CdsV:CdsO interface. Mutations of L638 and D639 of CdsV c to alanine abrogate binding between CdsV c and CdsO 25-110 , as measured by isothermal titration calorimetry (S10A and S10B Fig). The CdsV C : CdsO 25-110 complex has a Kd of 28 ± 3 μM, whereas the L638A/D639A mutant does not appear to bind CdsO (S10A and S10B Fig). D639 forms a salt bridge with H51 of CdsO   (Figs 2C and 4B), which may account for the importance of this interaction. These mutations have a minimal effect on stability as WT CdsVc and L638A/D639A have melting temperatures of 58˚C and 55˚C, respectively (S10C Fig). These residues are located within a broadly conserved region in SctV proteins and are invariant between C. pneumoniae and Pseudomonas aeruginosa (Fig 4A and S4 Fig). The homologous mutations, L635A/D636A, were made in Pseudomonas aeruginosa (pcrD) and bacteria were evaluated for secretion competency. Presence of the effector proteins ExoT and ExoS, and translocator proteins PopB and PopD, in Pseudomonas aeruginosa PA01 ΔexsE culture medium was compared with the presence of secreted proteins in the L635A/D636A double mutant and in wild type PcrD ( Fig 4C). As expected, ExoS and ExoT were detected in supernatant from WT PA01 ΔexsE containing the

PLOS PATHOGENS
Structure of the Type III secretion system export gate with CdsO Ca 2+ chelator EGTA, while PopB and PopD were detected in both the presence and absence of Ca 2+ (Fig 4C) [33]. The L635A/D636A double mutant was partially defective for secretion. We note that mutant PcrD was also expressed at a lower level than an unmutated epitope tagged control (Fig 4C), such that while we cautiously suggest that the contribution residues 635 and 636 make to the PcrD-PcsO interaction is important for maintaining secretion, the reduced secretion could be due to an unrelated stability affect in PcrD that is not seen in CdsV.

Discussion
Our structure of the CdsV C : CdsO 25-110 complex provides, for the first time, molecular details of the interaction of an export gate apparatus with the central stalk protein of a T3SS. Two new findings stem from this structure. First, the structure reveals that CdsO binds in an inter-subunit cleft between subdomain 4 of adjacent protomers, rather than in the central pore of CdsV (Fig 2). This region borders the recently described binding site, between subdomains 2 and 4 of a single protomer, for chaperone-cargo complexes (Fig 3 and [18]). Second, we observed

PLOS PATHOGENS
Structure of the Type III secretion system export gate with CdsO full occupancy of the CdsV C binding sites by CdsO. The 9:9 stoichiometry of the CdsV C :CdsO interaction observed in our structure indicate that symmetric binding is possible; however, as others have shown, only one CdsO may bind the ATPase at a time. Finally, the CdsV C :CdsO 25-110 complex allows a structural interpretation of mutations in SctV and SctO proteins that have been shown to alter secretion kinetics.
The CdsV C :CdsO 25-110 structure shows a symmetric 9:9 stoichiometry, as expected for a symmetric nonamer. It is also clear that only one CdsO can bind the ATPase at a given time [25]. The significance of the 9:9 stoichiometry may be that the ATPase bound CdsO can be directed toward any of the 9 available binding sites on CdsV. The modest affinity between CdsO and CdsV c (28 μM) suggests that CdsV likely interacts in vivo with ATPase-bound CdsO. While no direct measurement for CdsO concentration in Chlamydia is known, it is not among the~470 relatively abundant proteins assessed by quantitative mass spectrometry, and is likely less abundant than CdsV, which was observed [34]. We suggest that chaperone binding may increase the modest affinity between CdsV and CdsO such that the ATPase-bound CdsO preferentially engages CdsV already loaded with chaperone-cargo complexes. SctO proteins may promote cargo delivery by bridging the central "pore" of T3SS ATPases to the periphery of the export gate. The EscN:EscO cryo-EM structure shows a single EscO protruding from the asymmetric EscN hexamer at a~70˚angle [25], while CdsO 25-110 exits the CdsV C ring at an~60˚angle (S9E Fig). As shown in the EscN:EscO complex, lysines and arginines of the central stalk EscO directly interact with glutamate residues lining the pore of the ATPase EscN, which, concomitant with ATP hydrolysis, likely provide the rotational force of the ATPase to twist CdsO [25]. Given that, in the EscN:EscO structure, the two helices of the EscO coiled-coil are relatively equal in length, it is unknown how far a single helix of the central stalk may penetrate the ATPase in cases such as for CdsO, wherein the central stalk is asymmetric and, in general, longer than EscO. For F 1 -and V 1 -ATPase complexes, the central stalk extends around 70 Å into the catalytic core [20,21]. Manual modeling of an extended CdsO structure easily bridges the gap between the export gate platform and ATPase seen in the tomographic reconstruction from Salmonella (Fig 5 and [10]), with an additional~50 Å situated within the density assigned to the ATPase (Fig 5C). Thus, both structures support SctO proteins connecting the ATPase pore with the periphery of the export gate.
Comparison of the structures of CdsV C and the CdsV C : CdsO 25-110 complex indicates that binding of CdsO to CdsV alters the adjacent binding site for chaperone-cargo complexes. We suggest, based on the rotational movement expected from the EscN-EscO structure [25], that the ATPase-catalyzed twisting of CdsO could release chaperone-cargo complexes. This would release substrates from the export gate by disrupting the SD2-4 cleft and might also create a pathway between subunits to the secretion pore. The FlhA:FliS and FlhA:FliT structures [18] show cargo binding to the periphery of the export gate such that the secretion is initiated by cargo entering the export gate from the periphery, which could be initiated by the ATPase twisting CdsO.
Reports of direct interactions of the export gate and ATPase complex have included the observation that the interface between FlhA and FliJ (CdsV and CdsO homologs) is mediated by conserved residues Phe72 and Leu76 of FliJ (S9C Fig). Mutations of these residues significantly reduced FliJ's binding affinity for FlhA [31]. These residues instead likely serve to stabilize the FliJ coiled-coil. Manual docking of FliJ into the CdsO binding site of CdsV C indicates that the closest sidechain, F72, is >6 Å from CdsV C and pointing back toward the hydrophobic core of the FliJ coiled-coil (S9C Fig). Conversely, mutants within the same region of PscO, the CdsO homolog of Pseudomonas aeruginosa, upregulated secretion [35]. However, these residues lie lower along the PscO coiled-coil than the interaction interface that we have observed in our structure (S9D Fig). Additionally, mutation of several residues of FlhA have been shown to inhibit binding to FliJ, including FlhA residues E351, D356, R391, K392, K393, and

PLOS PATHOGENS
Structure of the Type III secretion system export gate with CdsO L401 [19]. These residues align to subdomain 1 of CdsV C ; thus, they are not directly involved in binding CdsO.
In summary, we present the atomic features and interaction interface of two critical components of the T3SS cytoplasmic sorting complex-the export gate CdsV with the ATPase's central stalk protein CdsO. Suprisingly, CdsO does not engage the central pore of CdsV, but instead docks at a peripheral intersubunit interface and is positioned to create an opening between CdsV subunits allowing a route for bound cargo to enter the secretion apparatus. Further biophysical studies will be essential to describe how the energetics of ATP hydrolysis and the proton motive force are coupled to drive contraction and dilation of the export gate to promote virulence.

Expression and purification of CdsV C and CdsO
CdsV residues 345-710 was amplified from Chlamydia pneumoniae and cloned into a pET28 expression vector, to utilize the vector's N-terminal hexa-His-tag and thrombin cleavage site. Protein expression was performed at low temperature (18˚C for 16 hours) in BL21 Star (DE3), after addition of 1 mM isopropyl β-d-thiogalactopyranoside for induction. Bacteria were collected by centrifugation and flash frozen in liquid nitrogen for later use. Bacteria were lysed with an Emulsiflex homogenizer (Avestin) in 25 mM sodium phosphate pH 8.0, 150 mM NaCl, with 10 μg/mL leupeptin, 1 μg/mL hen egg white lysozyme, 1 mM PMSF, 1 μg/mL DNase I and 0.7 μg/mL pepstatin. The lysate was clarified by centrifugation, and CdsV C was purified with Talon metal affinity resin followed by gel-filtration in 10 mM HEPES pH 7.5, 150 mM NaCl (when proteins were prepared for ITC, 500mM NaCl was used). Nonameric peak

PLOS PATHOGENS
Structure of the Type III secretion system export gate with CdsO fractions (CdsV C is a mixture of monomer/dimer and nonomer during gel-filtration) were pooled and concentrated with an Amicon ultrafiltration cell to 2mg/ml for crystallization. The CdsV C L638/D639A mutant was purified following the same protocol as for the WT CdsV C .
CdsO residues 25-110 were also amplified from Chlamydia pneumoniae and cloned into pET28. Protein expression and purification were performed as described above for CdsV, with the exception that the final gel-filtration buffer contained 300 mM NaCl.

Crystallization and data collection
CdsV C was crystallized by hanging drop vapor diffusion from a reservoir containing 100 mM HEPES pH 6.75 and 5% polyethylene glycol-6000 (PEG-6000), at 21˚C. Crystals were obtained after~2 weeks, cryoprotected using crystallization buffer supplemented with 20% glycerol, and cryo-cooled in liquid nitrogen. For heavy atom derivates, crystals were soaked in 1 mM heavy atoms in mother liquor for 2 days and harvested as for native crystals. X-ray data were collected at 100 K at LS-CAT Sector 21 at the Advance Photon Source (Argonne, IL). The data-collection statistics are given in Table 1. Diffraction intensities were processed and scaled with XDS [36]. Crystals were relatively non-isomorphous and an AuCl 2 -soaked crystal, with no evidence of bound gold, was used as a native. The data obtained from the crystals soaked in three heavy atoms-AuCl 3 , PtCl 4 and UO 2 (CH 3 CO 2 ) 2 , as well as an AuCl 3 soaked "native" used as the input to SHARP [37] to solve the phase problem using Multiple Isomorphous Replacement (MIRAS). This led to the determination of 6 Pt-sites, 23 U-sites, and 27 Au-Sites by employing the MR-SAD program in Phenix [38,39]. Phasing and density modification using SHARP resulted in a clearly interpretable electron density map.
Crystallization CdsV-CdsO was performed using multiple CdsO constructs, with the final structure including CdsV C and residues 25-110 of CdsO (CdsO  ). CdsV C and CdsO  were mixed with~10% molar excess of CdsO  and crystals were grown from 100 mM Bis-Tris pH 6.0, 4% PEG 3350, and 200 mM ammonium acetate. The CdsV C -CdsO 25-110 structure was determined by molecular replacement using CdsV C in Phenix.

Structure determination and analysis
A partial model of CdsV was built in COOT [40] and used to identify non-crystallographic symmetry operators, which were then used in Phenix to perform further rounds of density modification. The complete structure was built using COOT, refined in Phenix, and evaluated against 2mFo-DFc and mFo-DFc maps calculated in Phenix. This structure was used as a search model with a non-isomorphous native dataset. Five percent of the reflections from all datasets were used for R free sets.

Biophysical measurements
Isothermal titration calorimetry measurements were performed on a TA Instruments Nano ITC. Measurements were performed at 20˚C with a 300 μL cell volume and 24 x 2 μL injections with a stirring rate of 150 rpm. All proteins were in 10 mM HEPES pH 7.5, 500 mM NaCl. The respective protein concentrations were 15.0 uM, 12.1 uM, and 2.2 mM for nonameric CdsVc,

PLOS PATHOGENS
Structure of the Type III secretion system export gate with CdsO nonameric CdsVc L638A/D639A, and monomeric CdsO  . Calculations were performed using the TA instrument ITC Analyze software.

Thermal unfolding
The stability of WT CdsV C and the CdsV C L638/D639A double mutant were assessed by thermal unfolding based on intrinsic tryptophan fluorescence. The WT and mutant CdsV C proteins were diluted to 50 μM with gel filtration buffer and loaded into Tycho NT.6 capillaries (NanoTemper Technologies, Germany). Experiments were performed using a NanoTemper Tycho NT.6 instrument. The temperature gradient monitored was from 35 to 95˚C, increasing by 0.5˚C sec -1 . Protein unfolding was recorded by measuring changes in tryptophan fluorescence at emission wavelengths of 330 and 350 nm as a function of temperature. Inflection temperatures were determined by automatic fitting of fluorescence ratios (350/330 nm) with a polynomial function, where the maximum slope corresponds to the peak of its first derivative.

Pseudomonas secretion assay
PAO1F ΔexsE, PAO1 ΔexsE pcrD-VG2 and PAO1 ΔexsE pcrD(L635A+D636A)-VG2 were grown in LB supplemented with 2.5 g/L NaCl to late log phase. Cultures were harvested and resuspended in 2 mL LB with or without 5 mM EGTA. After 30 min, 1 mL of culture was pelleted, and protein was precipitated from 500 uL of supernatant. The pellets were resuspended and normalized to a final OD 600 of 2.5. Samples were separated by SDS-PAGE on a 10% gel (BioRad) and transferred to a PVDF membrane. With the exception of RpoA and VSV-Gtagged PcrD, the indicated proteins were detected by Western Blot using affinity purified rabbit antisera. RpoA was detected using a commercial mouse monoclonal antibody (BioLegend), and VSV-G using a commercial rabbit antibody (Thermo).
Supporting information S1 Table. Residues modeled into electron density in the CdsV C and CdsV C :CdsO structures. (DOCX)

PLOS PATHOGENS
Structure of the Type III secretion system export gate with CdsO

PLOS PATHOGENS
Structure of the Type III secretion system export gate with CdsO Salmonella (P0A1K2), Vibrio (A0A0H6WY40), and Escherichia (B7UMA5). Thr 67, the residue at the center of the loop connecting the two helices of the CdsO coiled-coil, is indicated with a black star. (B) Representative cartoon of CdsO determined in this work, colored according to sequence conservation (using ConSurf), with an extended model of CdsO, shown as both cartoon and surface representation. More conserved residues are located near the N-and C-termini of the coiled-coil. (C) An extended model of CdsO, colored by electrostatic potential (red is negative, blue is positive). The two views are obtained by 180˚rotation about the y-axis.  [31]); FliJ from Salmonella (orange; 3AJW; [1]); and YscO from Vibrio (purple; 4MH6). (C) The structure of FliJ manually docked into the CdsO binding site of CdsV, with FliJ residues proposed to influence export gate binding and secretion [48] shown as sticks. (D) The structure of YscO manually docked into the CdsO binding site of CdsV, with residues that impact secretion shown as sticks [35]. The corresponding residues from PscO are in parentheses. (E) Angles between the extension of CdsO from CdsV C and EscO from EscN from the crystal structure and cryo-EM structures (6NJP), respectively, are noted.

PLOS PATHOGENS
Structure of the Type III secretion system export gate with CdsO